Skip to content Skip to sidebar Skip to footer

Action and Automatism

Editor’s note: the debate between Diarmuid Costello and Charles Palermo on photography first began in the Summer 2012 issue of Critical Inquiry and continues on their website under the heading Debating Photography.


I am going to abstract, here, from the ideas of “automatic” and “automaticity,” and from “mechanism” and “mechanicity,” in order to focus on the following question: how should the idea of “automatism” be understood, and what theoretical resources can be used to illuminate it? This is what is at stake between Palermo and me. Palermo understands automatism as a peculiar species of action. Whether it can be legitimately considered a kind of action at all depends on how the notion of automatism itself is cashed out. Depending on that, I suggest, the answer will be “yes,” “maybe,” or “no.”


1. Varieties of Automatism

In the original debate between us, on which this exchange draws, Palermo appeals to the OED to show just how broad the meanings of “automatism” are when viewed in a broader historical perspective than recent theory of photography allows. The historical evidence Palermo adduces is compelling; it is whether it is also internally consistent that exercises me. What all the varieties of automatism that Palermo canvasses have in common is that they pick out acts that are not self-consciously thematized as such by the agent, yet which are acts performed by that agent nonetheless. In his examples Palermo goes well beyond the “two kinds of automatism” (habitual working methods on the one hand, and the free associations of the unconscious that such methods are supposed to free up on the other) that Rosalind Krauss employs to illuminate William Kentridge’s working procedures. It is Krauss on Kentridge that is the original source of our debate: in defending Krauss’s bipartite account from some criticisms I raise in passing, Palermo introduces, without thematizing or perhaps even being fully aware of the fact, a tripartite model of his own.1 Schematically:


  1. actions performed mechanically or unthinkingly, whether because habitual, conventional or rendered “second nature” by training;
  2. actions performed independently of conscious control, such as free association or automatic writing;
  3. actions unavailable to consciousness, because they take place at the level of motor or neural processes that subtend reflective awareness.


For the sake of brevity I shall describe these here as “preconscious,” “unconscious” and “unavailable to consciousness” respectively, though I decline the full theoretical burden that these terms may entrain in this or that psychoanalytic theory of mind. So let me state that by “preconscious” I mean any knowledge a subject possesses that is not presently conscious, but that can be called to mind or consciously attended to by an act of will; by “unconscious” I mean beliefs, attitudes or knowledge not presently conscious but in principle capable of becoming conscious, though not necessarily through an act of will; and by “unavailable to consciousness” I have in mind mental or bodily processes that are in principle incapable of becoming conscious. Note, first, that there is a difference of kind rather than degree between the first two and the third: that is, between sub-sentient processes that could not be reflectively thematized, because they take place at a motor or neural level that subtends‚ and so is inaccessible to, reflective awareness, and thoughts and activities that are merely no longer or not yet thematized—whether because conventional, habitual or even repressed, though they are in principle capable of being thematized. Call such (putative) “actions” strongly and weakly automatic respectively. Claims about motor or neural processes that we could not be aware of belong to the former; claims about what we are not (as things stand) aware of, though we could in principle be brought to awareness of, belong to the latter. This is not to deny that there is also a significant difference within the latter class between what I can attend to through an act of will, and what may be unavailable volitionally: it is to maintain that in so far as I could in principle become aware of such thoughts, beliefs or emotions, this a difference of degree rather than kind. Though Palermo refrains from discussing what I call strongly automatic actions here, in our original debate he endorses all three as bona fide varieties of automatism, and seeks to ground them in Anscombe’s theory of action. So what, one might think.



2. Anscombe on Action; Palermo on Automatism

The problem is that, on Anscombe’s account, strongly automatic actions could not properly be regarded as actions at all, because they fail her test of bona fide actions. As something we could not know about, being below the level of possible reflective thematization, they are not susceptible to the question ‘Why?’ in the relevant, reason giving sense that Anscombe has in mind.2 One cannot meaningfully ask someone why they are doing something they could have no awareness of doing. On Anscombe’s account, for something to count as an act there must be at least one description of that act such that, as so described, the agent can be intelligibly asked for their reasons for doing it: the act, in Anscombe’s classic and influential formula, must be “intentional under some description.”3 This is compatible with the same act being unintentional under a wide range of other descriptions. Thus the agent can be said to do many things of which they have no awareness, in doing whatever it is that they are aware of doing. A tennis player, for example, can be meaningfully asked why he tried a cross-court pass to his opponent’s backhand side rather than a driving down the line, but not why he was creating little eddies in the air or swatting midges, both of which may equally describe the bare act of swinging his racket head, but neither of which are descriptions under which that act is rationally endorsed by the agent. Or rather, one can ask this, but in so far as the agent is likely to reply that they were unaware of doing either, the act will not count as intentional under those descriptions. For just this reason agency cannot be reduced to intention. There are many things agents do without knowing that they do them; nobody else created those eddies or killed those midges. And they can be said to do those things, as opposed to those things befalling them, because there is something that they do know they are doing which involves the same act as those things that they do not know they are doing under various other descriptions. So intention does not exhaust the space of agency.

So far so good: but does this create room for all the kinds of case that Palermo calls “automatisms” and wants to construe as peculiar species of action? No. Take them in turn, starting with Palermo’s “weakly” automatic examples: habit, convention and so on may be straightforwardly accounted for in Anscombian terms. One may not reflectively attend to formulaic social greetings, habitual actions such as turning door handles to leave familiar rooms or reaching for tools that always occupy a given place on the tool board, or to the individual movements of one’s fingers in playing a piano piece one knows well, but one can be meaningfully asked them all the same. “Why did you greet him like that? “Why do you ask: it’s the decent thing to do!” “Why did you extend your right hand in that way without looking up?” “I wanted the hammer.” “Why did you move your fingers like that? “I was trying—but failing—to make that difficult transition.”

What about the harder to call intermediate cases, such as free association and other unconsciously determined processes? Here I think the most honest answer that can be given is “maybe.” In so far as a token belief, thought, emotion or action is unconscious now it will not count as intentional under any description. Ex hypothesi, if the agent is unaware of holding or doing it, they cannot be expected to say why they do. But saying that it would not count as intentional (under such circumstances) is much weaker than saying that it could not count as intentional. For it leaves open that whatever it is that the agent is doing or feeling could in principle become conscious and then re-described in ways that Anscombe’s account would capture. “I just can’t understand why you would say that!” “I didn’t want to acknowledge it at the time, but I think I was angry about…” versus “I’m sorry, I don’t understand it either, I just heard myself saying it.” Anscombe herself is noticeably equivocal about such cases, and perhaps this slipperiness explains why. She recognizes that the question “Why?” asked of something unfortunate blurted out in conversation is in order, but holds that the answer is that there is no answer.4 At this point, the psychoanalytically minded reader might want to add: at least so long as the reasons remain unconscious. Be that as it may, this seems to be an intermediate case: whether it will count as an action will depend on the circumstances of a particular case, and these may change over time.

By contrast, it seems to me that Anscombe’s account, properly understood, is incompatible with Palermo’s characterization of strongly automatic actions, and I now need to say why. Central to Palermo’s attempt to enlist Anscombe’s theory of action to characterize a wide variety automatisms as actions is the idea, reiterated here, of a “felt difference between a description and an intention,” or a difference between an “explicitly” and an “implicitly” intended action. The former would be any action that I can straightforwardly recognize as intentional under the description I might give of my own action. The latter is something that, if I understand Palermo correctly, I can somehow—Palermo declines to say how—acknowledge as my intentional act, in the sense that I grant that it is neither an unintended consequences of something that I did intend, nor an accident that befell me, but without being able to say how it relates to what I straightforwardly or explicitly intended. This is all very murky, and I believe that much of this murkiness has to do with the attempt to shoe horn the kind of cases that Palermo is interested in (dissociation, hypnosis and the like) into a theoretical framework that is clearly unsuited to their elaboration. On this theory such examples could not count as actions, at least as so described.

Such murkiness is only exacerbated when Palermo hones in on Anscombe’s account of the difference between acts that are and acts that are not “swallowed up” by broader descriptions that take into account a wider range of circumstances under which those acts, as narrowly described, take place. To adapt Palermo’s example of trying to take a well-focused picture: if I were to try to take a picture from the gently rolling deck of a ship at sea, then standing still, holding the camera steady, trying not to unwittingly obscure the lens (etc) would all be “swallowed up” by the description “trying to take a sharp picture of the rolling sea.” That is, I could be asked why I was standing still or holding the camera steady and the answer would be: because I am trying to get a sharp picture of the rolling swell. In so far as all of the component acts are done in the service of the act as more broadly described, the narrower descriptions will be “swallowed up” by that broader description. This means that all will count as intentional under that broader description, irrespective of whether or not I reflectively thematized any of them. None are unintended consequences or accidents. Whether or not I need to thematize them may depend largely on whether I am an amateur or a professional: think of how many things the experienced press photographer working at speed in dangerous environments needs to have mastered as automatisms to avoid being immediately killed. All such things, including being sensitive to what is about to explode into shot, will count as intentional under the description, as Robert Capa might have put it, of “getting the shot.” If this is what Palermo meant by “implicitly” as opposed to “explicitly” intended I would have no problem with either his account or his appeal to Anscombe to underwrite it. In so far as all such cases are intentional despite not being reflectively thematized, they fulfil Palermo’s most compelling descriptions of automatisms and confirm the value of appealing to William James’s account of habit as the progressive mastery of increasingly complex bodily acts (from walking to fencing or playing piano) as a way of unpacking automatism.5

The problem is that Palermo explicitly abjures this way of understanding an automatism, despite the fact that it is the natural way to understand much of what he does say, and the fact that his own examples of automatisms taking the form of high-level, well-drilled performance would seem to require it. For what is “implicit” as he puts it in his earlier piece, is supposed to characterize what is not “swallowed up” in this way by broader descriptions. This is an odd understanding of “implicit:” it entails that if x is “implicit” in y, then x is at odds with y. I don’t see how Palermo can have this both ways. Whatever is not swallowed up by a description is not intentional under that description—implicitly or otherwise—yet automatisms are supposed to count as implicitly intended under a description on Palermo’s account. The problem comes out most clearly in the examples that Palermo elaborates at greater length in our previous exchange. There Palermo maintains, consistent with his account here, that: “An automatic action is not an accidental one; it is an intentional operation. […] We call an action automatic if we acknowledge it as our act (as opposed, say, to an unintended consequence) but want to distinguish it from the description under which we perform that action” and, further, that “we want to speak of an automatic act as automatic if we are tempted to note some disconnection (or disconnections) between our intentions in performing it and the description under which we act.”6

The heart of our disagreement is this: if one wants to call an automatism an act—whether explicitly or implicitly intended—and one wants to enlist Anscombe in support of this account, then that automatism will have to count as intentional under the relevant description. Palermo appears to grant this. What he refuses to grant is in fact an entailment of this: that it will thereby be swallowed up by that description. On the theory of action that Palermo appeals to, there just is no “disconnection” between intentional action and description to be had. On the contrary, being “intentional under some description” is what makes a bodily movement an action; something is an action only if it is intentional under some description. Anything not swallowed up in this way does not contribute to the act in question; and if it does contribute to the act, it will be swallowed up by the relevant description. This is an either/or choice: if Palermo wants to maintain both that automatisms are actions and that they can be captured by the machinery which Anscombe’s account offers, then he has to accept the conceptual consequences of doing so. Failing that it looks like he will have to abjure appealing to Anscombe to underwrite the account he favours.


1. Diarmuid Costello, “Automat, Automatic, Automatism: Rosalind Krauss and Stanley Cavell on Photography and the Photographically Dependent Arts,” in Critical Inquiry 38.4, special issue “Agency and Automatism: Photography as Art since the Sixties,” Costello, Margaret Iversen and Joel Snyder, eds. (Summer 2012): 819-854; Rosalind Krauss, ‘“The Rock:” William Kentridge’s Drawings for Projection,’ October 92 (Spring 2000): 3-35.
2. G.E.M. Anscombe, Intention (Cambridge: Harvard University Press, 2000), §§ 5-7, 16-18.
3. Intention, §§23, 26 and passim.
4. Intention, §17.
5. “Habit” in William James, The Principles of Psychology, vol. 1 (New York: Dover, 1950), 104-127; especially, 108, 112, and 114-117.
6. Charles Palermo, “Automatism,” Critical Inquiry (forthcoming, 2014).